Application of machine vision in plant factories

Abstract

Plant factories have great potential for mitigating the contradiction between the growing global population and food scarcity. Machine vision plays an important role during automated food production, covering each production stage from raising seedlings, transplanting, management, harvesting and grading to packaging. In this paper, the prospects of machine vision application in plant factories are analyzed, then the present research utilizing this technology are summarized, including plant growth monitoring, robot operation assistance and fruit grading. Through the analyses, it is found that although existing methods have solved various practical problems with the advantages of low cost, high efficiency and high precision, there are still various challenges facing machine vision. Firstly, changing lighting conditions, complex indoor environments and color similarities between a plant’s organs can cause common image segmentation algorithms to fail. The lack of standard agricultural datasets hinders deep learning and unsupervised classification algorithms from making significant progress. Secondly, there are various theoretical knowledge gaps regarding the application of machine vision in the specific plant factory environment, which seriously hinders its application effect. Thirdly, a shortage of special image acquisition devices and supporting facilities has resulted in poor image quality. All these factors hinder machine vision to promote the development of plant factories. Nevertheless, this technology remains a powerful tool and is currently irreplaceable. We believe that machine vision will become more robust, efficient and reliable with the development of computer technology, the application of deep learning and updated algorithms.